Franchising, Starbucks vs. Subway, and Promotional Services

Josh Wright —  22 November 2007

Professor Bainbridge offers a correction to Keith’s Starbucks analysis by pointing out that Starbucks does not have franchisees. I don’t think the franchise/ franchisee distinction has much to do with Keith’s conclusion that whatever is going on is not an antitrust problem. But the Professor is on to a really cool question about franchising and vertical integration. Professor Bainbridge presents the contrasting franchising decisions by Starbucks and Subway as a transactions cost puzzle and links to a fuller analysis of this problem here:

What bugs me about the Subway v. Starbucks problem is that I can’t see any reason to believe that Subway’s transaction cost schedule is going to differ from that of Starbucks.

Bainbridge continues with the conventional account of the economics of franchising, focusing primarily on monitoring costs:

Franchising gives a residual claimant-like status to the local franchisee, while the franchise contract gives the franchisee incentives to ensure that the local employees comply with brand requirements. Franchising thus can be understood as an adaptive response to the problem of monitoring numerous employees in countless locations.

If this analysis is correct, one would expect to see corporate ownership in settings where monitoring via a vertically integrated management structure can be effected at low cost (relative to situations in which franchising dominates). But does Starbucks really face lower monitoring costs than Subway? If not, did Starbucks make an economic error by not going the franchise route?

The comments to Professor Bainbridge’s prior analysis from the folks at Marginal Revolution (and others) following the post hint at cross-store cannibalization as the key to the economic explanation. This is basically right, or at least, on the right track. But I think all this talk about “monitoring” and “transactions costs” is covering up some really interesting economics that shed some light on vertical contracting more generally. Don’t get me wrong, the critical variable that should influence the vertical integration decision here is indeed (as Professor Bainbridge says) the monitoring costs of franchisor owned outlets relative to franchised oulets. But I think these rather vague labels are blurring some important underlying economics.

So what’s going on with Starbucks’ decision to vertically integrate its outlets? I offer some analysis below the fold.

I. The Incentive Incompatibility Between Franchisors and Franchisees

Like most vertical contracts, the key here is to understand how the incentives of the prospective transacting parties do not coincide and therefore must be controlled contractually rather than left to unrestrained competition and self-interest. There are a number of well understood malincentive problems in the franchise context. For example, Rubin (1978) and Klein (1980) examined the free-riding problems associated with common use of a brand-name, e.g. a reduction in quality by the franchisee reduces demand facing all franchisees using the common brand name and therefore the individual franchisee has a lower incentive to supply optimal quality.

A second example is Telser’s (1960) “special services” explanation of vertical restraints wherein the contractual restraints are designed to “solve” the problem of customers consuming pre-purchase services at the full-service outlet before purchasing at the discount dealer (I’ve discussed this at length in a prior post in the context on Leegin and RPM). A third is the also well understood example of solving double marginalization or “successive monopoly” problems.

Another common incentive incompatibility, identified by Klein & Murphy (1988) and later analyzed by Klein (1995 and elsewhere), occurs when: (1) franchisors (manufacturers) sell a product at a significant markup over marginal cost, (2) the franchisee (retailer) provides some input like marketing activity or promotion that has a significant impact on demand for the product, and (3) . This “promotional services” incompatibility is much more common than the other three, and in my view, has far greater explanatory power in terms of understanding modern uses of vertical restraints. The basic economic forces under these conditions suggest that the downstream “promotional service provider” such as a franchisee or retailer does not have adequate incentives to promote the product or supply the efficient level of marketing activity. This is because the franchisee does not take into account the franchisor’s (large) profit margin on additional sales induced by provision of promotional services. This is most likely to be the case when products are differentiated, i.e. automobiles, soft drinks, and in today’s world … coffee! Anybody know what the the marginal cost of a vanilla latte is? Anyway, this point was first raised in Klein and Murphy (1988) and is distinct from the Telser “special services” explanation.

Under these conditions, transacting parties will find contractual solutions to these problems (including vertical integration) to induce the supply of the efficient level of promotional services. My analysis with Ben Klein on slotting contracts (forthcoming JLE) and solo authored work on category management contracts are examples of the types of contracts one sees put to use in the retail industry to control the transacting parties incentives in favor of non-performance and faciliate self-enforcement of the contract.

Obviously, these contracts require monitoring of difficult to measure franchisee/retailer performance in terms of the supply of marketing activity and promotional effort. The obvious implication for vertical integration is the monitoring costs of franchisor owned relative to franchisee outlets. In particular, when franchisee malincentives are very large and the franchisor is unable to write a contract directly on desired franchisee behavior or to pay the large required premium, vertical integration may be the cheapest way for the franchisor to assure the supply of desired distribution services.

II. What Promotional Services? Or, So What Does Any of This Have to Do With Starbucks?

Starbucks doesn’t rely on promotional shelf space like my slotting contract example, and for that matter, neither does Subway. So why the big difference in contractual form? The answer lies is understanding the nature of promotion. This is where Tyler and Alex correctly observe that Starbucks’ outlet density is very high relative to other franchises. There’s an excellent reason for this. The optimal density of outlets is different precisely because for some products, the number of outlets in a given area may influence product demand for some consumers. “Impulse” purchases are a good way to think about this. For some consumers, the additional convenience of having outlets on every block might induce them to purchase when they would not have at some lower density level. In essence, the number of outlets in a form of promotion or marketing activity. However, an increase in outlet density will also reduce the demand for nearby franchisees.

This is where Tyler and Alex’s comments on Starbucks ignoring cross-store cannibalization in favor of building the “brand name” come in. This is somewhat correct, but again, blurs some of the underlying economics. Notice, for example, that even if the externality between new and old franchisees were internalized, the “promotional services” incentive incompatibility would still exist! Klein (1995) discusses the case where the franchisee is granted the decision rights over new outlets in the territory and notes that the franchisor and franchisee incentives would still not be aligned precisely because the franchisee does not take into account the incremental profit earned by the franchisor from additional sales. In other words, the franchisor might gain from additional outlets even though the extra sales at the retail level might not cover the cost of the extra outlets. It is the franchising contract that must solve the problem of inadequate outlets (or more generally, franchisee promotion).

III. Revisiting Starbucks vs. Subway

So, if you’ve made it this far I figure I better give my answer for the difference between Starbucks and Subway even though the analysis above more or less spells it out:

  1. Starbucks’ demand is more sensitive to outlet density than Subway because it relies more significantly on “impulse sales”
  2. Starbucks incremental profit margins are likely higher than Subway (just a guess on this one — no data)
  3. (1) and (2) mean that the incentive incompatibility between the Starbucks and its would be franchisors is much larger than for Subway.
  4. While it would still be possible for Starbucks to contract to solve this incentive problem, such contracts would require some assurance to the franchisee regarding changes in outlet density. These is a particularly difficult dimension to contract over because what Starbucks really wants is the flexibility to change outlet density with demand conditions and negotiate the premium it pays to its outlets accordingly.
  5. Contracting over outlet density is likely to create rigidity that Starbucks would like to avoid.

Therefore, one should expect that attempts to solve this promotional services problem would include more vertical integration for Starbucks than Subway. Vertical integration does not eliminate the possiblity of non-performance in the supply of valuable promotional services that translate to profitable impulse sales, but it does eliminate some aspects of the non-performance problem.

Muris, Scheffman & Spiller (1992) provide a similar analysis of the shift to vertical integration in the soft drink distribution market following a dramatic increase in the importance of marketing activity in the industry, e.g. supplying retailers with product display, “pushing” product by encouraging retailers to give premium shelf space with “slotting contracts,” and executing local promotions. It is true that one could call this change in optimal contractual form as a response to increasing transactions costs, but that is probably a bit misleading. Most folks assume that this means a response to an increased incentive to engage in hold up over specialized assets. But this incentive to vertically integrate has nothing to do with specialized assets in the conventional Klein, Crawford, and Alchian (1978) or Williamsonian sense.

As an aside, I believe a failure to understand the role of vertical restraints in facilitating self-enforcement of contractual relationships (including vertical integration) involving promotional services has caused much mischief in antitrust law.


6 responses to Franchising, Starbucks vs. Subway, and Promotional Services


    As for the transaction cost and monitoring explanation for Starbucks vs Subway contracting differences, they are probably valid, however probably not in the way as discussed by that air force guy in 2003 in a comment on Bainbridge’s blog.

    In matters of decentralization, the location of knowledge matters. Processing food relies much more on delicate local knowledge than serving a cup of coffee. Whether it is about the ingredients and the way of processing (monitoring), or whether ingredients can be bought locally or centrally (supply). Also knowledge of customers’ preferences can be crucial in a competitive environment (demand), this all affects food more than coffee.

    Having worked in a restaurant myself, I know that serving coffee is highly standardized, and requires little knowledge (or extended decision rights). The monitor of a Starbucks employee would be bored immediately as the entire process is automated: one kilo of ground coffee beans gives you 50 cups of coffee. Any hick-ups are the result of poor machine maintenance, or sour milk. There is little room for ambiguity.


    In addition to the comments above, Starbucks coffee is hardly addictive, believe me as a Continental European. However, when people want coffee, they want it now, here. Consumers drink coffee as a habit, no surprises, please. Even with high margins, coffee remains a commodity.

    People are willing to pay for outlet density, especially if the outlet offers cheap space to sit and work without the obligation to consume. That is any franchise holder’s nightmare: the loitering client. Such clients are likely not to be found at Subway.



    Yes, the repeat sales certainly matter. In my analysis, they would “fit in” to the extent that they increased the gap between franchisor and franchisee profitability associated with a marginal increase in promotional output. That impact on profitability is, I would argue, very consistent with this framework.


    Very interesting, Josh.

    Might there also be a store-to-store externality associated with the creation and maintenance of an addiction? If Starbucks coffee is more addictive than Subway products, then perhaps there’s a greater positive externality among Starbucks stores than there is among Subways, which may then be one factor that could help explain why Starbucks has a relatively higher outlet density than Subway. Needless to say, this is not to discount but rather to supplement the other factors that you identify as possibly being at work.

Trackbacks and Pingbacks:

  1. TRUTH ON THE MARKET » Starbucks, Subway, and Antitrust - November 23, 2007

    […] THE MARKET » Starbucks, Subway, and Antitrust on Starbucks Store Traffic and Nonexclusivity.Ideoblog on Franchising, Starbucks vs. Subway, and Promotional Services.Josh on Franchising, Starbucks vs. Subway, and Promotional Services.Keith Sharfman on Franchising, […]

  2. Ideoblog - November 23, 2007

    Why firms franchise: Starbucks vs. Subway

    Steve Bainbridge and Josh Wright post on the title question. I won’t summarize the posts or replicate the many links, particularly including links to an earlier 2003 discussion. Here I focus on my paper with Jon Klick and Bruce Kobayashi,